Dying is not what it used to be! Impact of evolving epidemiology and treatment on mode of death in heart failure

How will a patient with heart failure (HF) die? With an aging population, increasing medical complexity, and evolving therapeutic options, clinicians may find it harder to answer this question. While tools, scores, and calculators have been deployed to estimate overall prognosis, cause-specific mortality is less frequently considered. The accurate and updated accounting of how patients die, however, has important implications for their overall outlook, healthcare decisions, and treatment priorities. Across disease states, these estimates may change dynamically with age, co-morbidities, and attendant therapies.

In HF with reduced ejection fraction (HFrEF), there has been welcome and substantial scientific progress in disease management over the last several decades. Randomized clinical trials have demonstrated the salutary benefits of several therapeutic classes on major modes of death, including death related to progressive worsening HF and sudden cardiac death.¹ Broad implementation of these therapies in clinical practice would be expected to lead to meaningful changes in the expected prognosis of patients with HFrEF. In this issue of the Journal, the article by Moliner et al.² examining longitudinal shifts in prevailing modes of death allows us to reflect on this recent therapeutic progress and to look ahead at emerging challenges for patients with HFrEF.

Contextualizing changing patterns of death in heart failure with reduced ejection fraction

Moliner et al.² examine changes in patterns of death among 1876 patients recently hospitalized for HFrEF from 2001 to 2018 who were followed in a structured multidisciplinary clinic in Spain. Patients referred were comparable to contemporary clinical trial populations of HFrEF with respect to age (64.8 vs. 65.8 years), gender distribution (men: 77% vs. 75%), New York Heart Association class (II or III: 95% vs. 93%), and ejection fraction (average 28% vs. 31%), respectively.¹

Over a median of 4.2 years, there were 935 deaths of known cause, accounting for ~50% of the cohort. While most deaths were cardiovascular across nearly all years of follow-up, there was a significant decrease in the proportion of cardiovascular deaths over time, primarily driven by a decline in rates of sudden death. Notably, rates of HF death did not change substantially over time. Conversely, there was a significant increase in non-cardiovascular deaths over time, which were most commonly due to cancer or infection. Among cancer deaths, lung cancer was most common.

The investigators should be congratulated in sharing detailed adjudicated mode of death data of patients followed by a stable HF referral network over a couple decades with very limited loss to follow-up (n = 11). It is important to contextualize these observed temporal trends. First, few patients (~15%) with HF and a preserved ejection fraction were referred to this structured HF clinic and these patients were not considered in this analysis. Second, patients with unknown death (7%) were excluded from the analysis. Undetermined deaths are variably handled by clinical event committees in clinical trials³; however, HFrEF trial programmes have traditionally ascribed these deaths as cardiovascular in nature, an assumption rooted in historical mode of death distribution. Sensitivity analysis classifying these undetermined deaths as of cardiovascular cause did not appreciably change the study findings. Third, the rates of mechanical circulatory support or referral for heart transplantation were not reported and may impact causes of death. Fourth, Spain and other high-income countries have experienced population-level aging during the study time frame.⁴ As patients who died of non-cardiovascular causes were on average older than patients who died of cardiovascular causes, these observations may be partially related to era effects and variation in ages of cohorts. It is thus challenging to discern if a HFrEF patient in 2001 may truly be comparable to the HFrEF patient in 2018 with respect to age, co-morbidities, and healthcare structures of care.

Implications of declining sudden death in heart failure with reduced ejection fraction

The current study adds to the growing body of evidence surrounding the decreasing rates of sudden death among patients with HFrEF. The definition of sudden death in the current investigation was similar to that used in contemporary HF clinical trials. Despite the known issues regarding the specificity of such definitions to identify arrhythmic sudden death when compared with a gold standard of autopsy,⁵ the adjudication process was rigorously performed by a dedicated clinical events committee. Various evidence-based medical therapies for HFrEF possess mechanistic effects that may prevent malignant dysrhythmias and have shown to decrease rates of sudden death in pivotal outcomes trials.¹ Sacubitril/valsartan, which was prescribed to a minority of patients in the current investigation, has been demonstrated to reduce sudden death compared with enalapril,⁶ and thus rates of sudden death in HFrEF may continue to decline with expanding use of this therapy in the community. Lack of autopsies in contemporary medical practice however makes it impossible to differentiate arrhythmic from non-arrhythmic sudden death.

The decline in rates of sudden death carries important implications regarding patient candidacy for implantable cardioverter-defibrillators (ICDs). It is not known whether the decline in sudden death is related to increased utilization of ICDs over the study period or due to increased adherence to HF medical therapies. However, utilization of ICD therapy was low in this population (10–12%) and therefore it is difficult to ascribe the decline in sudden death to ICDs alone. These results suggest that contemporary guidelines that recommend a 3-month ‘waiting period’⁷ to facilitate initiation and titration of medical therapy prior to re-evaluation of ejection fraction to determine candidacy for ICD therapy may be insufficient. As rates of sudden death are declining and initiation and dose adjustments of evidence-based therapies have proven to be suboptimal in real-world settings,⁸ a 3-month period may be inadequate, and may potentially lead clinicians to offer ICDs to HFrEF patients who may not ultimately succumb to sudden death or who may experience improvement in their ejection fraction to a range where ICD therapy is not indicated.

Broad implementation of ICD therapy for primary prevention of sudden death in HFrEF may be less applicable, especially among cohorts who face high risks of non-sudden death.⁹ There is a continued need to identify patients at high risk for arrhythmic death specifically, who serve to benefit most from ICD therapy. Indeed, risk scores that incorporate clinical variables that are associated with disproportionate risk for sudden cardiac death compared with all-cause mortality, including younger age, male sex, and high body mass index, have proven useful.¹⁰ Moving forward, certain imaging techniques may offer promise and warrant further study Speckle-tracking echocardiography has identified several variables, including global longitudinal strain and indices of mechanical dispersion, that are associated with death to due ventricular arrhythmia.¹¹ Additionally, myocardial scar assessment by cardiac magnetic resonance is a strong predictor of malignant arrhythmia.¹² Finally, biomarker-based strategies for risk stratification could be considered in the future, but current biomarkers including natriuretic peptides do not sufficiently differentiate risk of arrhythmic vs non-arrhythmic death.

The rise of non-cardiovascular deaths among heart failure patients

Hidden in these data is the unsettling increase in overall number of deaths in this Spanish cohort, driven by increases in non-cardiovascular deaths over time, while the absolute number of cardiovascular deaths have not appreciably changed. Recent increases in HF-related mortality within the US from 2013 to 2017 have also been reported.¹³ Among the general population, Spain and several other high-income countries have had recent unexpected declines in life expectancy for men and women. While these adverse trends have been driven by external factors such as drug use in the US, excess cardiovascular and respiratory causes may account for recent observed declines in life expectancy in Spain.⁴

Cancer appears to account for most of the recent increase in non-cardiovascular deaths in this cohort. The increased lifespan among those who died of non-cardiovascular causes compared with those who experienced cardiovascular deaths may be in part related to the protective benefits of evidence-based HF therapies. Furthermore, increased recognition and broader availability of screening for various cancers may lend to lead time bias, and increased adjudication of cancer-related deaths. While it is certainly possible that the presence of HF increases the future risk of cancer, it appears more likely that patients with mild forms of HF who were effectively treated may die on account of competing causes of death. Indeed, among a cohort of > 25 000 male physicians, HF was not associated with either cancer incidence or cancer-specific mortality.¹⁴ As the current data call for further investigation into the presence of a potential mechanistic relationship between HF and cancer, they also offer the opportunity to reflect on the need for collaborative care of HF patients (Figure 1). A multidisciplinary care framework is crucial, which may include age-specific cancer screening in addition to routine healthcare maintenance measures. As many patients contend with multiple, potentially intersecting serious illnesses, early involvement of palliative care and hospice support systems may be needed.¹⁵

Figure 1.

Implications of changes in mode of death among contemporary heart failure with reduced ejection fraction (HFrEF) patients. HF, heart failure; ICD, implantable cardioverter-defibrillator.

Periodic re-examination of major causes of death in HFrEF determines if prior targets of therapy are truly being modified and identifies future therapeutic challenges. Reduction in sudden death in clinical practice reflects success in the development and use of effective guideline-directed medical therapies. HF-related mortality, a traditional endpoint and target of therapy, does not appear to be durably modified in practice. New therapies in the therapeutic pipeline with direct myocardial effects offer hope to alter this dominant cardiovascular mode of death. Observations of increases in non-cardiovascular deaths re-emphasize the growing burden of co-morbidities in an aging population and the need for innovation in healthcare delivery models centred around holistic and comprehensive care for patients living with HFrEF.